Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/sheawinkler/contextlattice/agents-mdgit clone --depth 1 https://github.com/sheawinkler/ContextLatticeWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/sheawinkler/contextlattice/agents-md)<a href="https://agentmods.dev/instructions/sheawinkler/contextlattice/agents-md"><img src="https://agentmods.dev/badge/instructions/sheawinkler/contextlattice/agents-md.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00687 | $0.00687 |
| Opus 5 | $0.00344 | $0.00344 |
| Sonnet 5 | $0.00137 | $0.00137 |
| Haiku 4.5 | $0.00069 | $0.00069 |
Grade A, and why
ContextLattice AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 43 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Instructions
ContextLattice routes local memory and capabilities for this repo.
Start
- Use
http://127.0.0.1:8075; set bothCONTEXTLATTICE_ORCHESTRATOR_URLandMEMMCP_ORCHESTRATOR_URLto it. - Before non-trivial work, run
contextlattice_agent_start --soft --compactorscripts/agent_hooks/agent_start.sh --soft --compact. - For agent-native sessions, prefer
contextlattice_agent_adapter bootstrap --agent <profile> --project <project>to track runtime state, handoffs, and reusable prompts.
Work Loop
- Retrieve scoped context before broad reasoning:
- narrow:
POST /memory/search - broad/multi-file:
POST /memory/context-pack - helper:
scripts/agent/contextlattice-pack "<task>" --project contextlattice - session helper:
scripts/agent/contextlattice-session context-package --session-id <session_id> --pretty
- narrow:
- Read only returned files unless blocked or incomplete.
- Prefer repo-local conventions and deterministic scripts over repeated inference.
- Run checks that match the changed surface.
- Write durable decisions, failures, commands, and changed assumptions back:
scripts/agent/writeback --topic-path <path> --file <logical.md> --kind <kind> --stdin
- Before a major handoff or problem-solving prompt, run
contextlattice_agent_session context-package --session-id <session_id> --prettyand use the bounded result as the factual scaffold.
Hooks
- Use hooks for runtime health, git lane, env drift, rebuild, leak, resource, recall, and checkpoint checks.
- Keep judgment here: correctness, safety, maintainability, minimal context, and repo convention beat style preference.
- Do not paste long runbooks, API docs, examples, or architecture lore into always-loaded instructions. Put them in references/docs or ContextLattice.
Host Lifecycle Safety
- Host supervisors, installers, schedulers, runtime start/stop paths, and recovery policy are host-lifecycle critical.
- Before merge, complete the PR evidence contract; run
scripts/agent/audit-host-supervisor-safety, matching failure injection, an installed upgrade smoke under an isolated scheduler identity, and at least two real scheduled intervals with identity/restart-count proof. - Source tests do not prove installed behavior. Never reuse an operator scheduler label or tag a host-lifecycle change without tested rollback.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 43 lines · 687 tokens per session scan A 43aea2721370
ContextLattice AGENTS.md is an instructions file published in the GitHub repository sheawinkler/ContextLattice (153 stars, last pushed 9d ago), licensed Apache-2.0. It adds 687 tokens to every session, about $0.0034 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
memory-journal-mcp copilot-instructions.md
Copilot instructions for neverinfamous/memory-journal-mcp, covering memory journal mcp — copilot code review context, project overview, session context, tooling standards and coding standards.
aictx AGENTS.md
Instructions for oldskultxo/aictx, covering agents instructions for aictx and ai context engine.
aictx aictx.instructions.md
Instructions for oldskultxo/aictx: Apply these instructions to repository work. Prefer AICTX MCP tools when available; fall back to AICTX CLI commands otherwise. They intentionally duplicate the minimal AICTX lifecycle because Copilot support varies by surface.
structured-context-spec CLAUDE.md
Instructions for tim-mccrimmon/structured-context-spec, covering claude.md, structured context specification (scs) — project context, what this project is, repository structure and tools & development commands.
openwebui-extensions copilot-instructions.md
Copilot instructions for Fu-Jie/openwebui-extensions, covering copilot instructions for openwebui-extensions, 🏗️ 项目结构与命名 (project structure & naming), 1. 语言与代码规范 (language & code requirements), 2. 插件目录结构 (plugin directory structure) and 3. 文档字符串规范 (docstring standard).
Commit Message
Comprehensive Conventional Commits guidelines for openwebui-extensions.